USGS1975
Networks of small stream (drainage area less than 50 square miles or 130 square kilometres) flood gages have been operated throughout the country for a number of years to supplement flood information already available for large streams. The goal in operating these networks has been to obtain sufficient data for estimating flood frequency at ungaged sites with the equivalent accuracy expected from 10 years of observed flood records. In some areas the networks have accumulated sufficient data to satisfy these accuracy goals. A review of these networks, looking toward possible reduction of the number of gages, is now timely. Continued operation of a few selected gages may be desirable to provide a longer time-sample base for improving the flood-frequency estimating equation and(or) to expand the area over which the equations apply. In 1974, Thomas Maddock III developed a rational method for selecting gages to be retained in a reduced hydrologic network. This method of network analysis seeks the optimum set of gages to be retained for a given level of annual operating costs with the information content of the reduced network being the factor optimized. Application of Maddock's method demonstrated that a considerable number of gages could be eliminated from a network without grossly decreasing its information content. Maddock's method of analysis is described in detail for a hypothetical network of gages. The method also is applied to actual networks in Montana, Illinois, and Georgia. The analysis of networks in Montana illustrates the basic approach to selecting an optimal subset from the existing set of gages. The Illinois analysis demonstrated that by retaining only 26 percent of the gages, nearly 63 percent of the original information is retained in the reduced network. Application of the procedure shows how the design of networks in Georgia may be modified because of hydrologic considerations to meet budgetary constraints.
Water-Resources Investigations Report↗